AI-assisted data engineering: production-grade pipelines in weeks, not quarters.
The audit is the low-risk way in. Most larger projects should start there.
Scopes are fixed, so the work has an edge.
Retainer hours are capped in the agreement.
These ranges are hypotheses to test on the first sales calls. Balkan clients will likely sit at the low end, and remote EU or US clients higher. A call turns a range into an agreed scope and a price.
Data audit
€1,200–3,000 fixed
1–2 weeks
You know the reporting is messy and you want a written plan before paying for a larger build.
You get
A map of the systems that hold the numbers you report
Where data quality slips, and where reports disagree
The reporting gaps worth closing
A written roadmap of what to build and what to leave alone
Pipeline and warehouse
€4,000–12,000 fixed
3–6 weeks
You have two to five sources — project management, a CRM, a shop, ads, or similar — and no single place the numbers come from.
You get
Ingestion of those sources into Postgres or BigQuery
Models on top of the warehouse
One to three dashboards
AI reporting assistant
€3,000–8,000 pilot
2–4 week pilot
You already have a cleaned model, and people want to ask it questions without writing SQL.
You get
A text-to-SQL assistant on the model you approve
Guardrails so questions stay on that model
A pilot with a clear end, rather than an open-ended build
Monthly retainer
€600–2,000 per month
Monthly
A pipeline is already in place and you want it kept up without an open-ended commitment.
You get
Monitoring and fixes
Small changes inside the cap
New reports, as the hours allow
An hour cap written into the agreement
Questions that come up
Where should a project start?
With the audit, unless a warehouse and a trusted model are already there. The roadmap from the audit is what a pipeline or an assistant should be priced against.
Why a fixed scope instead of an hourly rate?
The work has a written edge: what is in, how long it runs, and what done means. A retainer adds an hour cap so a month cannot grow on its own.
Is the number on this page the invoice?
No. Each figure is a range to test. The price for a given piece of work is fixed once the scope is agreed, on a call.
Where does the work happen?
Async, from Skopje. Slovenia and the Balkans come first, then remote teams in the EU and the US. We do not need to share an office.
What does the warehouse run on?
Postgres or BigQuery. Models and one to three dashboards are part of the pipeline package. The way data is pulled in depends on the sources, and it is named in the proposal.
Can people ask questions of the data?
That is the assistant pilot: text-to-SQL on a cleaned model, with guardrails. It sits on top of a model you already trust. It is not a way to skip the cleanup.